Cognitive Computing is a rapidly developing technology that has reached practical application and implementation. So what is it? Do you need it? How can it benefit your business?
In this webinar a panel of experts in Cognitive Computing will discuss the technology, the current practical applications, and where this technology is going. The discussion will start with a review of a recent survey produced by DATAVERSITY on how Cognitive Computing is currently understood by your peers. The panel will also review many components of the technology including:
Cognitive Analytics
Machine Learning
Deep Learning
Reasoning
And next generation artificial intelligence (AI)
And get involved in the discussion with your own questions to present to the panel.
2. 2
A panel of experts will discuss the technology, where it’s headed and current practical
applications. The discussion will start with some slides from recent DATAVERSITY on
how Cognitive Computing is currently understood by your peers.
The panel will also review many components of the technology including:
• Cognitive Analytics
• Machine Learning and Deep Learning
• Reasoning and next generation artificial intelligence (AI)
• And get involved in the discussion with your own questions to present to the panel.
All webinar registrants will be sent soon to be published Research Paper on Cognitive
Computing produced by DATAVERSITY and co-authored by Moderator Steve Ardire.
Included in the paper is as a coupon code to receive a $200 discount on first annual
Cognitive Computing Forum to be held in San Jose, California August 20 – 21st.
The following charts represent responses to
DATAVERSITY Survey on Cognitive Computing
conducted in May and June 2014. All data, charts, and
analysis contained in these slides and report remain the
copyright of DATAVERSITY Education, LLC.
About the Webinar
3. Steve Ardire
‘Merchant of Light’
sardire@gmail.com
@sardire www.linkedin.com/in/sardire/
Steve advises cognitive computing, AI, machine learning startups
and for past 20+ yrs advised / consulted with 30+ software
startups in US, Canada, Europe in semantic technology, Big Data,
cloud computing, predictive analytics, infoviz, DAM, plus much
more
3
Moderator
4. 4
Panelists
Steve advises
learning startu
consulted with
Europe in sem
computing, pre
much more
www.linkedin.
Tony Sarris Founder and Principal, N2Semantics
tony.sarris@n2semantics.com
www.linkedin.com/in/tonysarris
@n2semantics Technology evangelist, consultant and advisor specializing in
semantic technologies (e.g., knowledge representation, ontology, intelligent software
agents). Tony would like to see apps that are software agents or automated
assistants actually helping us humans in our day-to-day lives.
James Kobielus Big Data Evangelist, IBM
jgkobiel@us.ibm.com
@jameskobielus ( FOLLOWERS 13K )
www.linkedin.com/pub/james-kobielus/6/ab2/8b0
Industry veteran and IBM Big Data Evangelist; Sr Prog Dir, Product Mktg, Big
Data Analytics; Editor-in-Chief, IBM Data Mag. He spearheads thought leadership
activities in Big Data, Hadoop, enterprise data warehousing, advanced analytics,
business intelligence, data management.
Adrian Bowles Founder of STORM Insights, inc.
adrian@storminsights.com
www.linkedin.com/in/ajbowles
@ajbowles STORM Insights - a new approach to market intelligence
Adrian is an industry analyst and recovering academic, providing research and
advisory services for buyers, sellers, and investors in emerging technology
markets including cognitive computing, big data/analytics, and cloud computing.
He has held executive positions at several consulting and analyst firms.
5. 5
Some Highlights from Cognitive Computing Survey
• 53.4% of respondents believe that Cognitive System technologies need to provide more
clarity in terms of business perspectives.
• 16.7% of respondents either are not aware of technologies like IBM’s Watson, Siri, and
Google Now or don’t find them applicable to a discussion of Cognitive Computing.
• Some of the most needed resources for eventual enterprise integration of Cognitive
Systems include better education of benefits, more case studies, easier to use tool sets,
and vendor demonstrations.
• More than a third of respondents said that they were still unclear about their organization’s
plans for Cognitive Systems implementation due to a lack of understanding about how to
present the business case.
• 26% remarked that their organizations are early adopters of emerging technologies such
as Cognitive Computing, NoSQL, and Big Data.
• Almost 60% said that one of the primary drawbacks to current integration with Cognitive
Systems is the lack of knowledge and skills among existing IT staff, especially in terms of
Data Scientists and Machine Learning experts.
• Business Intelligence/Cognitive Analytics was the top choice (81.8%) for how Cognitive
Computing can help the enterprise.
9. 9
Application Engagement and UI
( Insights, UIX experience )
Descriptive, Predictive, Prescriptive, Cognitive Analytics
algorithms, AI / Machine Learning ( Supervised and Unsupervised ) Deep Neural
Network Learning
ETL with NLP text analytics, entity extraction etc
Structured Unstructured Streaming Other
Big Data, Smart Data,
KnowledgeBase
( SQL, NoSQL, RDF or combinations of ) ecosystems becoming enterprise data hubs
Data Sources
refine &
improve
refine &
improve
refine &
improve
by Steve Ardire @sardire
Similarities in Big Data & Cognitive Computing Stack
12. 12
IMO the 3 key differentiators of Cognitive
Computing Solutions
( your thoughts)
1) Context driven dynamic algorithms for automating
pattern discovery and knowledge.
2) Reasons and learns instantly and incrementally to
discern context for sense-making.
3) Cognitive Systems infer, hypothesize, adapt, and
improve over time without direct programming.
13. 13
Jan 30, 2014 By James Kobielus As big data analytics pushes deeper into cognitive
computing, it needs to bring the Semantic Web into the heart of this new age
Cognitive computing can't achieve its potential without a strong semantic-processing
substrate that executes across diverse content sources.
You could view cognition in the cloud as the next evolutionary plateau for the
semantic Web but this demands we have a shared understanding of the relationship
between the concepts of "cognitive computing" and "semantic computing."
James please elaborate on Cognitive computing can
take the Semantic Web to the next level
15. 15
In How to say AI ( It’s a long “A” and Short “I” ) was trying to
illustrate in my own post about the complexity of encoding
knowledge. Knowledge representation is not the same as text
processing, or natural language processing for that matter.
Tony Sarris @n2semantics · Feb 23
Concerns about #MachineLearning http://goo.gl/tGybSK . How
about a little more 'designed serendipity’ ? http://goo.gl/FzfM6k
Tony please elaborate on your tweets below on
knowledge representation and designed serendipity
16. 16
Classification
Regression
Clustering
Collaborative Filtering
Category Algorithm Goal
Logistic Regression &
Random Decision Forest
Generalized Linear Models
K-means++
Alternating Least Squares
SupervisedUnsupervised
Pattern Recognition
Predict Future Values
Segment Historic Data
Recommend Items
“Google is not really a search company. It’s a machine-learning company”
- Matthew Zeiler, CEO of visual search startup Clarifai | Enterprise | WIRED
Let’s discuss Machine Learning ( the New Black )
Machine learning ( ML ) is branch of AI where algorithms process data, draw conclusion
17. 17
The State of Apache Spark in 2014
http://databricks.com/blog/2014/07/18/the-state-of-apache-spark-in-2014.html
Every major Hadoop distributor has made Spark part of their distribution with Spark replacing
MapReduce for Hadoop and other data stores (e.g. Cassandra, MongoDB)
Enterprises are deploying Spark for ETL, machine learning, data product creation, and
complex event processing with streaming data and increasingly is the backend for higher-level
business applications like for advanced analytics using Spark’s scalable machine learning
library (MLlib)
Vertical-specific cases include churn analysis, fraud detection, risk analytics, 360-degree
customer views.
18. 18
http://graphlab.com/products/create/overview.html
Democratization of machine learning ?
most machine learning research has focused on one-off systems and “my
curve is better than your curve” demonstrations but GraphLab thinks it can
reach 80 to 90 percent of use cases handling the full lifecycle from data
engineering through production without requiring a specialized, deep
engagement before it’s deployed.
( comments )
19. 19
Ayasdi | Automatic Insight Discovery $30.6M Series B July 2013 (Total $51M)
BeyondCore $9M Series A Feb 2014
BigML - Machine Learning Made Easy closed $1.4M seed
Context Relevant $21M Series B May 2014
Emerald Logic - closing $2M Series A
ersatz labs - deep neural networks in cloud closed $500K seed
Lumiata $4M Series A Jan 2014
Microsoft Azure Machine Learning - recently launched and is significant
Nervana Systems $600K seed April 2014
Nutonian, Inc. $4M Series A Oct 2013
0xdata H20 $1.7M in Jan 2013
PredictionIO Open Source Machine Learning Server just closed $2.5M seed
SKYMIND : DEEP LEARNING FOR EVERYONE
Skytree closed $18M Series A April 2013
Vicarious http://vicarious.com/ $40 Series B Mar 2014 (Total $60M)
Wise.io | Machine Learning as a Service $2.6M Series A Mar. 2014
VC $$$ is flowing to ML startups ( partial list )
20. 20
AUTOMATING THE DATA SCIENTIST by @louisdorard
http://www.louisdorard.com/blog/automating-the-data-
scientist?utm_content=bufferea381&utm_medium=social&utm_source=twitter.com&utm_campaign=buf
fer
To me, a Data Scientist is someone who has business acumen,
technical skills, and who knows Machine Learning. The last point is
key. Those who do not have Machine Learning expertise are Data
Engineers, Data Analysts, Data Artisans, Data Somethingelse.
Automating the Data Scientist
( comments )
21. 21
http://www.fastcodesign.com/3029756/why-algorithms-are-the-next-star-designers
• Rather than having a team of data scientists creating algorithms to understand
a particular business issue, cognitive analytics seeks to extract content,
embed it into semantic models, discover hypotheses and interpret evidence,
provide potential insights—and then continuously improve them.
• The data scientist’s job is to empower the cognitive tool, providing guidance,
coaching, feedback, and new inputs along the way. As a tool moves closer to
being able to replicate the human thought process, answers come more
promptly and with greater consistency.
Why Algorithms are the Next Star Designers
( comments )
22. 22
Artificial Intelligence Is Now Telling Doctors
How to Treat You
Vinod Khosla has 3 predictions for the future of health. We’ve got 1 more
http://venturebeat.com/2014/05/20/vinod-khosla-has-3-predictions-for-the-future-of-health-weve-got-1-more
1. 80 percent of what doctors do, diagnostics, will be replaced by machines
2. Medicine will become tailor-made for each patient
3. Consumer-driven tech will create better incentives to keep people healthy
4. We’ll move from disease care to actual health care thanks to pervasive monitoring
23. 23
Let’s discuss Deep Learning
Deep learning is a set of algorithms that uses multi-layer neural networks that can teach
themselves to understand complex patterns, non-linear transformations, and features that
comprise the data they’re on which they’re trained.
Google, Facebook, Microsoft ( Project Adam ), Apple, et al embracing more powerful form of
AI known as “deep learning” to improve everything from speech recognition and language
translation to computer vision, the ability to identify images without human help.
Google improved Android’s voice recognition and acquired DeepMind for $400 - 500M,
Microsoft created live voice translation system called Skype Translate, Baidu has invested
$300M and picked up Stanford’s Andrew Ng, and Apple is building out a team too.
• Computers recently matched humans ( correct 97.53% ) at facial recognition.
• Facebook DeepFace 97.25% required 7.4 million images for training its system,
• Chinese University of Hong Kong 99.15% accuracy with 200,000 images using a better classifier and
neural network (essentially a vast artificial brain)
With deep learning, computer scientists build software models that simulate—to a certain
extent—the learning model of the human brain e.g. After nine years of research, Numenta
finally has apps that mimic the way the brain works (interview)
http://venturebeat.com/2014/07/09/numentas-brain-research-has-taken-a-long-nine-years-but-it-starting-to-
pay-off-interview/
24. 24
Cognitive Computing UI’s contingent on vertical use case and
targeted user e.g. clinician, knowledge worker, consumer
Life Sciences
Oil & Gas
Public Sector
Financial Services
Manufacturing
Retail
Collaboration
Customer Service
etc.
Data Shape Has Meaning
26. 26
• Watson now has 800 clients and partners, like Sloane-Kettering Cancer Center
for suggestions on treating cancer patients.
• Watson used as a back end for very specific interfaces IBM has created
IBM’s billion-dollar bet on Watson
July 18, 2014 http://venturebeat.com/2014/07/18/inside-ibms-billion-dollar-bet-on-watson/
was big news last week
Siri as a ‘general assistant’ could provide lots of data to improve Watson's intelligence
so perhaps 27 years later Apple Knowledge Navigator
concept https://www.youtube.com/watch?v=umJsITGzXd0
can be made real with @ibmwatson
27. 27
Cognitive Computing competition is intensifying
Google to develop "fully reasoning" AI but real-life
Skynet still a few years off per co-founder Sergey Brin
http://www.pcmag.me/a/2460571
IBM recently
acquired AI
startup Cognea to
give personality
to virtual personal
assistants and to
better understand
personality of
users
Her name is Cortana. Her attitude is
almost human. http://engt.co/1nesb9H
Microsoft's decision to infuse Cortana
with a personality stemmed from one
end goal: user attachment. "We did
some research and found that people are
more likely to interact with [AI] when it
feels more human," said Susan
Hendrich, project manager in charge of
overseeing Cortana's personality
28. 28
Take ‘Personality’ to another level with psychobiological simulation which
learns and interacts in real time with neuroscience models
Emotions are Data so imagine a machine that can laugh and cry, learn, dream, and express its
inner responses to how it perceives you to feel with emotional intelligence.
The Laboratory for Animate Technologies at Univ of Auckland
http://www.abi.auckland.ac.nz/en/about/our-research/animate-technologies.html is combining
Bioengineering, Neuroscience, AI and Interactive Computer Graphics to define next generation of
human computer interaction.
BabyX incorporates multiple learning models including unsupervised learning, reinforcement
learning, conditioning and action discovery using a specialized framework called Brain Language
29. 29
Robotic Personal Assistants
Edmonton airport is testing out customer
service robots designed to interact with
people, as well as detect and display
emotions. The robots can not only give you
directions, but actually take you where you
need to go. And they have the potential to
do so in 30 different languages.
Meet JIBO, The World’s First
Family Robot. Friendly, helpful and
intelligent. JIBO is the real deal,
from social robots pioneer Cynthia
Breazeal. JIBO can’t wait to meet
you.
http://www.myjibo.com/
30. 30
Neuromorphic computers and
Neurosynaptic chips
Like a brain, a neural processing unit (NPU) processes many different data
streams at the same time. The end goal is to have devices that can read
complex sensory information (like voices) at a fraction of the computational
cost of traditional chips.
This means that Siri’s daughter will be able to answer your questions faster,
with less prompting, and without being as much of a drain on your battery.
These NPUs will run alongside traditional, binary CPUs, which will still be
essential for running things like operating systems and tip calculators.
Intel, IBM, Qualcomm et al have neurosynaptic chipset programs to mimic the
behavior of the human brain and will be used for machine learning and
cognitive computing systems like Watson.
Sandia Laboratories working on neuromorphic computers that mimics human
brain both in terms of parallel processing power and efficiency.
32. 32
• Hong Kong venture capital firm just named an AI tool known as VITAL to its Board of
Directors with goal of finding better investments through more innovative decision making.
• Google using Cognitive Computing to improve the efficiency of its data centers.
• Spotting rare diseases in family photos from pictures of people with genetic disorders,
including Down's syndrome, fragile X syndrome and progeria
• MIT researchers are working on a activity-recognition algorithm that is learning how to
understand what is happening in videos, thereby eventually allowing the tagging of
indexing of vast online video collections.
• From solar panels to batteries, algorithms are becoming key to designing new materials
• Emerald Logic has created FACET (Fast Collection Evolution Technology) that tests tens
of thousands of algorithms to discover the most predictive and valuable ones.
• Allen Institute and Univ of Washington developing LEVAN (Learn Everything about
Anything) that can teach itself essentially “everything it needs to know” by examining
search engines through natural language processing and Machine Learning techniques.
Cognitive Computing Potpourri for $100
33. 33
Cognitive Systems hypothesize, recommend, adapt to learn from
interactions then reason through dynamic experience just like humans.
But it’s not about replacing humans with machines. It’s about
harnessing combined strengths of both to solve complex problems
from ever-changing factors and new information.
The “programmable era" of computers invariably will be transcended
by Cognitive Computing systems.
Final Observations